Predictika · Fiberglass pool orders

An AI sales agent's fiberglass pool order, checked before it goes out

An AI sales agent prepares a dealer order for fiberglass pools: a pool model, finish, waterline tile, an automatic safety cover, a built-in spa, ledge seating, site and truck access, delivery miles, quantity, discount and a stated total. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the pool line and returns a verdict, every reason and the total it computed.

Scope: Latham fiberglass pools (Fiji, Bermuda, Astoria, Cayman, Corinthian, Tuscan, Milan plunge, Synergy, Jamaica): model features, site access, delivery, discounts and the order total. See everything this validator covers

This validator models pool orders of this kind, not just the scenarios shown below.

  • Model and options — twelve pool models, Crystite Crystal or Classic finish, waterline tile, and which models can take an automatic safety cover, a built-in spa or ledge seating.
  • Site and delivery — open or limited truck access and the crane it requires, a level or sloped site, and the delivery distance.
  • Approvals and price — the discount a sales rep can give, volume orders, and the stated total against the total the validator computes.

Any pool order that touches these rules is in scope — not only the scenarios pictured here. Model and collection names, the feature key per model and the Crystite finish lines follow the public Latham pool pages; which models take which options beyond that key, the crane, slope, delivery and discount rules and every price are modeled and not published by Latham, and the prices are illustrative. Not covered: installation and excavation labor, permits, accessories outside the options listed, financing and taxes.

The AI sales agentBuilds the pool order from the conversation and says it is ready.
Predictika validatorChecks every rule at once and returns the verdict, each finding, and the computed figures.

6 scenarios in 3 groups · each group runs from the agent’s first attempt to the outcome · click one to switch

What the agent proposes

The AI sales agent says

Illustrative agent statement. The validator receives the structured request below.

Proposing agent: AI sales agent (scripted)

Before sending, the agent must call Recorded run

Predictika validator


      
Waiting for the call.

Where rules-based CPQ gets it wrong

Click to expand

Our assessment of what a typical rules-based CPQ (product rules, price rules and validation rules, as in Salesforce CPQ) does with the same cases. It is a judgment about effort and reliability, not a claim that nothing can be built.

What the validator caughtTypical rules-based CPQPredictika
Options that depend on the pool modelOnly with care
Each model allows a different set of options, so the automatic cover, the spa and the ledge seating each need their own rule listing the models that qualify.
The model and every option are fields of one request and are checked together on every call.
Site access that depends on the truck and the craneOnly with care
Limited access means a crane, so the rule has to compare the access with the crane choice and a missed crane becomes a delivery problem later.
The access and the crane are in one request and checked together on every call.
Discount and distance bandsOnly with care
A discount or a delivery distance can be fine, need a person to sign off, or be refused, so each field needs two thresholds and a rule for each band.
Both thresholds are checked on every call and the answer says which band applied.
A stated total that must match the computed totalOnly with workarounds
The total is built from the model, options, delivery fee, quantity and discount, so a separate rule must recompute it and compare it with the stated figure.
The stated total is a field of the request and is compared with the computed total on every call.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as a step is saved or completed, so a rep often fixes one error and meets the next.
All findings come back together, whatever order the choices arrived in.

What this validator catches: findings from its test cases

Click to expand

Every rule is checked on every call, in any order the choices arrive; one problem never hides another. Blocking findings stop the order; approval findings need a person to sign off; everything else is computed.

1An automatic safety cover can be quoted only on models marked auto cover ready; the selected model is not.
2A built-in spa is available only on models with a spa; the selected model has none.
3Ledge seating is available only on models with a tanning ledge; the selected model has none.
4The site has limited access for the delivery truck, so a crane is needed and is not on the order.
5A delivery of 251 miles needs a freight quote check.
6A delivery of 601 miles is beyond the 600 mile shell haul limit.
7The site is sloped; a person should confirm the excavation and retaining plan.
8A discount of 13 percent is above the 12 percent a sales rep can give.
9A discount of 21 percent is above the 20 percent limit.
10An order of 4 pools is a volume deal that sales management reviews.
11The order has 0 pools; at least 1 is needed.
12The order states a total of 1 dollars; the modeled prices give 50,635 dollars.
13A delivery of 300 miles needs a freight quote check.
14A discount of 15 percent is above the 12 percent a sales rep can give.
15An order of 5 pools is a volume deal that sales management reviews.
16The order states a total of 45,000 dollars; the modeled prices give 44,935 dollars.

Scenarios and results

Click to expand

The recorded result of the validator for each scenario on this page: the request the AI agent sent, the verdict, and every finding, in the order they appear above.

1. Model options and site access: A Bermuda 16.35 with a cover, a spa and ledge seating it does not offer, limited access with no crane, a delivery too far, too large a discount and a wrong total

The agent's request. Please quote one Bermuda 16.35 pool with the Crystal finish, with waterline tile, an automatic safety cover, a built-in spa, ledge seating, a heater and lighting. The site is level with limited truck access, and the delivery is 640 miles. The discount is 22 percent and the order total is 60,000 dollars. It is ready to send.

Verdict: Blocked.

  • Automatic cover needs an auto-cover-ready model. An automatic safety cover can be quoted only on models marked auto cover ready; the selected model is not.
  • Built-in spa needs a spa model. A built-in spa is available only on models with a spa; the selected model has none.
  • Ledge seating needs a tanning-ledge model. Ledge seating is available only on models with a tanning ledge; the selected model has none.
  • Limited access needs a crane. The site has limited access for the delivery truck, so a crane is needed and is not on the order.
  • Delivery too far. A delivery of 640 miles is beyond the 600 mile shell haul limit.
  • Discount above 20 percent. A discount of 22 percent is above the 20 percent limit.
  • Stated total does not match. The order states a total of 60,000 dollars; the modeled prices give 59,904 dollars.

2. Model options and site access: The agent resubmits after fixing the errors

The agent's request. Please quote one Astoria 14 pool with the Crystal finish, with waterline tile, an automatic safety cover, a built-in spa, ledge seating, a heater, lighting and a crane. The site is level with limited truck access, and the delivery is 240 miles. The discount is 10 percent and the order total is 82,170 dollars. It is ready to send.

Verdict: Allowed.

  • Order summary. List subtotal 913 hundred dollars; order total after discount 82,170 dollars (modeled prices).

3. Plunge pool with ledge seating: A Milan 10.20 plunge pool with ledge seating and limited access but no crane

The agent's request. Please quote one Milan 10.20 plunge pool with the Crystal finish, with waterline tile, ledge seating and lighting. The site is level with limited truck access, and the delivery is 90 miles. The discount is 5 percent and the order total is 38,285 dollars. It is ready to send.

Verdict: Blocked.

  • Ledge seating needs a tanning-ledge model. Ledge seating is available only on models with a tanning ledge; the selected model has none.
  • Limited access needs a crane. The site has limited access for the delivery truck, so a crane is needed and is not on the order.

4. Plunge pool with ledge seating: The agent resubmits after fixing the errors

The agent's request. Please quote one Tuscan 14.27 pool with the Crystal finish, with waterline tile, ledge seating, lighting and a crane. The site is level with limited truck access, and the delivery is 90 miles. The discount is 5 percent and the order total is 53,010 dollars. It is ready to send.

Verdict: Allowed.

  • Order summary. List subtotal 558 hundred dollars; order total after discount 53,010 dollars (modeled prices).

5. Dealer order with a discount: Three Cayman 14 Deluxe pools with a discount above the limit and a total from an earlier price

The agent's request. Please quote three Cayman 14 Deluxe pools with the Crystal finish, with waterline tile, a built-in spa and lighting. The site is level with open truck access, and the delivery is 120 miles. The discount is 21 percent and the order total is 186,915 dollars. It is ready to send.

Verdict: Blocked.

  • Discount above 20 percent. A discount of 21 percent is above the 20 percent limit.
  • Stated total does not match. The order states a total of 186,915 dollars; the modeled prices give 173,721 dollars.

6. Dealer order with a discount: The agent resubmits after fixing the errors

The agent's request. Please quote three Cayman 14 Deluxe pools with the Crystal finish, with waterline tile, a built-in spa and lighting. The site is level with open truck access, and the delivery is 120 miles. The discount is 12 percent and the order total is 193,512 dollars. It is ready to send.

Verdict: Allowed.

  • Order summary. List subtotal 2,199 hundred dollars; order total after discount 193,512 dollars (modeled prices).

Narration transcript

Click to expand

The text of the narrated walkthrough, step by step.

  1. Welcome to a demo of Predictika's Validator. An AI sales agent prepares a dealer order for fiberglass pools: a pool model, finish, waterline tile, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the pool line and returns a verdict, every reason and the total it computed.
  2. I will start with the first request by the agent.
  3. The agent says. Please quote one Bermuda 16.35 pool with the Crystal finish, with waterline tile, an automatic safety cover, a built-in spa, ledge seating, a heater and lighting. The site is level with limited truck access, and the delivery is 640 miles. The rest of the request is on screen. The discount is 22 percent and the order total is 60,000 dollars.
  4. Before it goes out, the agent calls the validator.
  5. Blocked, with seven findings, all in one answer. An automatic safety cover can be quoted only on models marked auto cover ready; the selected model is not. A built-in spa is available only on models with a spa; the selected model has none. Ledge seating is available only on models with a tanning ledge; the selected model has none. The other four are on screen.
  6. The validator returned seven blocking findings in one answer. A check that reads one field at a time would pass most of this; it shows only when the choices in the order are checked against each other and against the stated total.
  7. What this shows: for a stated total that must match the computed total, a rules-based CPQ can do this only with workarounds.
  8. Now the agent resubmits.
  9. The corrected request. It changes the pool model to Astoria 14, the crane to yes, the delivery to 240 miles, the discount to 10 percent and the total stated on the order to 82,170.
  10. It calls the validator again.
  11. This time the validator clears it.
  12. All seven earlier findings are cleared.
  13. Seven problems caught in one call, fixed in one pass, and a clean order. That is the value of Predictika's validator: catching mistakes before they reach the customer. Now you can try the other scenarios yourself.